Re: [Qemu-devel] [PATCH 01/13] hw/ppc.c: LOG_IRQ macro

[Eduardo Habkost]

On Thu, Dec 11, 2008 at 11:54:08PM +0100, Andreas Färber wrote:
>
> Am 11.12.2008 um 22:45 schrieb Anthony Liguori:
>
>> Eduardo Habkost wrote:
>>> This macro will avoid some #ifdefs in the code and create a single  
>>> point
>>> where the logging call can be changed in the future.
>>>
>>> Signed-off-by: Eduardo Habkost <address@hidden>
>>> ---
>>> hw/ppc.c |   10 ++++++++++
>>> 1 files changed, 10 insertions(+), 0 deletions(-)
>>>
>>> diff --git a/hw/ppc.c b/hw/ppc.c
>>> index 60d6e86..cbd69e0 100644
>>> --- a/hw/ppc.c
>>> +++ b/hw/ppc.c
>>> @@ -31,6 +31,16 @@
>>> //#define PPC_DEBUG_IRQ
>>> //#define PPC_DEBUG_TB
>>> +#ifdef PPC_DEBUG_IRQ
>>> +#  define LOG_IRQ(...) do {           \
>>> +     if (loglevel & CPU_LOG_INT)      \
>>> +       fprintf(logfile, __VA_ARGS__); \
>>> +   } while (0)
>>> +#else
>>> +#  define LOG_IRQ(...) do { } while (0)
>>> +#endif
>>>
>>
>> This style of macro is less ideal than:
>>
>> #define LOG_IRQ(fmt, ...) fprintf(logfile, fmt, ## __VA_ARGS__).
>
> Now that the TCG conversion is done, does QEMU still require GCC? I  
> remember we had issues with such macros in Mono's eGLib due to some  
> compilers needing "fmt..." instead of "fmt, ..." and some not supporting 
> "__VA_ARGS__" but "...".

I chose __VA_ARGS__ because I expected it to be more commonly supported on
other compilers. GCC documentation mentions "fmt..." as a GCC extension.


>
> It might make sense then to put such logging macros in a central place  
> (qemu-common.h?) to avoid having to cope with such annoyances in all  
> those source files.

Yes. I plan to unify all those debug macros somewhere, later. But I
want to do this one step at a time. This first series is just to keep
exactly the same behavior with a cleaner code. After that, we can make
all targets use similar macros and then unify them on a single place.

-- 
Eduardo